Jacobs Technion-Cornell Institute(雅各布斯以色列理工学院-康奈尔大学研究院);Cornell University(康奈尔大学);University of Toronto(多伦多大学)
Ronald J. Brachman, Hector J. Levesque
备注:
Initial version published in Proceedings of AAAI-22, the Thirty-Sixth AAAI Conference on Artificial Intelligence. Original version extended slightly to include acknowledgement of more recent work, including new references, and to clarify remarks in a few paragraphs
Common sense has always been of interest in Artificial Intelligence, but has rarely taken center stage. Despite its mention in one of John McCarthy's earliest papers and years of work by dedicated researchers, arguably no AI system with a serious amount of general common sense has ever emerged. Why is that? What's missing? Examples of AI systems' failures of common sense abound, and they point to AI's frequent focus on expertise as the cause. Those attempting to break the resulting brittleness barrier, even in the context of modern deep learning, have tended to invest their energy in large numbers of small bits of commonsense knowledge. While important, all the commonsense knowledge fragments in the world don't add up to a system that actually demonstrates common sense in a human-like way. We advocate examining common sense from a broader perspective than in the past. Common sense should be considered in the context of a full cognitive system with history, goals, desires, and drives, not just in isolated circumscribed examples. A fresh look is needed: common sense is worthy of its own dedicated scientific exploration.
Information Sciences Institute, University of Southern California(南加州大学信息科学研究所);Dept. of Informatics, Pontificia Universidade Católica Rio de Janeiro(里约热内卢天主教大学信息学系)
CUHK at SemEval-2020 Task 4: CommonSense Explanation, Reasoning and Prediction with Multi-task Learning
The Chinese University of Hong Kong(香港中文大学);Institute of Computing Technology, Chinese Academy of Sciences(中国科学院计算技术研究所);South China Normal University(华南师范大学)
本文描述了我们的系统提交给SemEval 2020的任务4:常识验证和解释(ComVE),其中包括三个子任务。基于BERT架构的多任务环境,提出了一个有效的、可解释的“解释、推理和预测”(Explain,Reason and Predict,ERP)系统来解决与常识相关的三个子任务:(a)验证、(b)推理和(c)解释。受常识认知研究的启发,我们的系统首先生成句子的原因或理解,然后选择哪一个陈述是有意义的,这是通过多任务学习实现的。在后评估中,我们的系统在子任务A中达到了92.9%的准确率(排名11),在子任务B中达到了89.7%的准确率(排名9),在子任务C中达到了12.9的BLEU得分(排名8)。
SemEval-2020 Task 4: Commonsense Validation and Explanation
Zhejiang University(浙江大学);Sun Yat-Sen University(中山大学);School of Engineering, Westlake University(西湖大学工学院);Singapore University of Technology and Design(新加坡科技设计大学)
CommonsenseQA: A Question Answering Challenge Targeting Commonsense Knowledge
Tel-Aviv University(特拉维夫大学);Allen Institute for Artificial Intelligence(艾伦人工智能研究所)
在回答问题时,除了特定的上下文之外,人们还经常利用他们丰富的世界知识。最近的工作主要集中在回答一些相关文件或背景下的问题,并要求很少的一般背景。为了研究具有先验知识的问答,我们提出了常识问答:一个具有挑战性的新数据集常识问答。为了捕捉关联之外的常识,我们从ConceptNet(Speer等人,2017)与单个源概念具有相同语义关系的多个目标概念。人群工作者被要求创作多项选择题,其中提到源概念,并依次区分每个目标概念。这鼓励工作人员创建具有复杂语义的问题,这些问题通常需要先验知识。我们通过这个过程创建了12,247个问题,并通过大量强基线证明了我们任务的难度。我们的最佳基线是基于BERT-large(Devlin et al.,2018年),并获得56%的准确率,远低于人类的表现,这是89%。
Breaking Common Sense: WHOOPS! A Vision-and-Language Benchmark of Synthetic and Compositional Images
Allen Institute for Artificial Intelligence(艾伦人工智能研究所);The Hebrew University of Jerusalem(耶路撒冷希伯来大学);Ben Gurion University of the Negev(内盖夫本古里安大学);University of Washington(华盛顿大学)
How to Understand Named Entities: Using Common Sense for News Captioning
School of Electrical and Information Engineering, Tianjin University(天津大学电气与信息工程学院);School of Computing, National University of Singapore(新加坡国立大学计算机学院)
SemEval-2024 Task 9: BRAINTEASER: A Novel Task Defying Common Sense
Information Sciences Institute, Viterbi School of Engineering, University of Southern California(南加州大学维特比工程学院信息科学研究所);Department of Computer Science, Faculty of Science, Vrije Universiteit Amsterdam(阿姆斯特丹自由大学理学院计算机科学系);Tencent AI Lab(腾讯AI实验室)